To elaborate our model we removed uncertain localities such as those localities with no coordinates and all records with coordinates laying down in sites with estimated elevations below 2,015 m and above 2,800 m. Also six records in BioMap four from Magdalena and two from Santander, which represent corrupted data in the database were deleted.
The habitat suitability model generated in Maxent showed a few sparse areas suitable in climatic terms for this species in the northern Western Andes, southern Central Andes and the east slope of the Eastern Andes in its central northern section. Areas not near the type locality or isolated from it were excluded from the potential distribution map of this hummingbird.
Assuming that the distribution of the species may have filled the complete climatic model generated, its distribution today in remnants of forest is about 645 km2, which corresponds to a loss of 38 % of its potential original distribution due to deforestation.
This species has been categorised as Critically Endangered (CR) according to BirdLife International (2016) because it is known from a few locations and has a very small area of occupancy in a region were still deforestation and fragmentation advance due to agriculture. BirdLife has estimated its extent of occurrence in 31 km2. Our model predict a larger extent of occurrence than the one estimated by BirdLife, being in the order of 1,049 km2, although sensu stricto just considering areas highly suitable this extent of occurrence is reduced to 233 km2. Most areas predicted as marginally suitable and suitable need confirmation of the presence of the species. Similarly to Eriocnemis isabellae, our maps constitute the first approach to understand better the geographical distribution of this rare species and can be used as a guide to direct searches of the species in other localities. In case it is proven the species may have a wider distribution possibly it can be in a near future down-listed to Endangered (EN). This species needs urgently research of its general ecology and distribution.
Regularized training gain is 5.215, training AUC is 0.999, unregularized training gain is 6.200.
Algorithm converged after 420 iterations (1 seconds).
The follow settings were used during the run:
5 presence records used for training.
10005 points used to determine the Maxent distribution (background points and presence points).
Environmental layers used (all continuous): bio10co bio11co bio12co bio13co bio14co bio15co bio16co bio17co bio18co bio19co bio1co bio2co bio3co bio4co bio5co bio6co bio7co bio8co bio9co
Regularization values: linear/quadratic/product: 1.000, categorical: 0.575, threshold: 1.950, hinge: 0.500
Feature types used: linear
'Equal Training Sensitivity and Specificity' and 'Equate Entropy of Thresholded and Original Distributions' thresholds and omission rates:
0.001-0.005-Fractional predicted area
0.000-0.000-Training omission rate